Supercomputer sharpens views from space telescope

Supercomputer sharpens views from space telescope

Sonet OC-3 link speeds raw images from observatories to supercomputing center, where a 32-node server digitally enhances

BY PATRICIA DAUKANTAS | GCN STAFF

The Air Force's Maui Space Surveillance System has boosted its computing capacity to identify fast-moving space debris while it is still in telescopic view.

The upgraded system does near-real-time image processing, said 1st Lt. Dale White, deputy program manager of the Maui High Performance Computing Center in Kihei, Hawaii.

The Defense Department-funded center receives electronic images transmitted from Air Force Research Laboratory telescopes on Haleakala, a nearby dormant volcano.

Processing power

The image processing happens on an IBM RS/6000 SP server with 32 four-processor nodes, 128G of memory and 1.15T of disk storage, said Steve Karwoski, associate director of operations. The IBM Power3-II processors have copper interconnects to cut power consumption.


align="right" width="210">

size="2" color="#FF0000">The Maui Space Surveillance Team on Haleakala mountain keeps watch for space debris that could harm aircraft and satellites.

If they need to, MHPCC staff members can cluster the SP with a classified, 48-node IBM SP, Karwoski said. Together, the 80 four-processor nodes provide theoretical peak performance of 480 billion operations per second.

The Maui center has more than 600 IBM nodes in all.

The Air Force lab operates six telescopes on Haleakala, which is also on the island of Maui. They capture images at a variety of visible and infrared wavelengths, White said. The largest of the six, the Advanced Electro-Optical System, has an aperture of 3.67 meters.

Air Force officials transferred control of the observatories from the Space Command to the laboratory last October, White said.

A Synchronous Optical Network OC-3 link speeds raw images from the observatories to the supercomputing center, where the 32-node SP digitally enhances them. Processing an image takes one to three seconds.

'That's pretty incredible when you think about it,' White said.

Improvement of the image-enhancing algorithms by researchers at the center and the University of New Mexico also helped shorten processing times, Karwoski said.

White said the Space Surveillance System captures and processes images requested by center customers'mainly Defense agencies, but also some civilian researchers.

Fender bender?

For example, the system can capture images of NASA's space shuttle in orbit to check for exterior damage incurred during launch, White said.

Scientists and military officials have a keen interest in tracking space junk'from small parts to dead satellites'that could damage spacecraft aloft.

The variety of telescopes and detectors in use makes it difficult to estimate the size of the smallest detail that the Space Surveillance Team can resolve, White said.

'We refine [each image] to the best level practical,' he said.

inside gcn

  • data science (chombosan/Shutterstock.com)

    4 steps to excellence in data analysis

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group